3834 papers
This paper introduces SpreadsheetArena, a platform for evaluating large language models' end-to-end spreadsheet generation capabilities through blind pairwise comparisons, revealing that while models can produce functional workbooks, they often fail to align with domain-specific best practices and that user preferences vary significantly across different use cases.
The paper introduces SENS-ASR, a streaming automatic speech recognition approach that improves transcription quality under low-latency constraints by injecting semantic information extracted from past frame-embeddings via a context module trained through knowledge distillation from a fine-tuned language model.
This paper demonstrates that fine-tuned bidirectional encoders, specifically a hybrid AraBERTv2 architecture, significantly outperform large-scale causal decoders in the challenging task of 82-class Arabic medical text classification by better capturing global semantic context despite data imbalance and label noise.
This paper introduces Personalized Group Relative Policy Optimization (P-GRPO), a novel framework that improves alignment with diverse individual preferences by decoupling advantage estimation from batch statistics and normalizing rewards against preference-group-specific histories, thereby overcoming the limitations of standard GRPO in handling heterogeneous user signals.
The paper introduces FERRET, a multi-faceted automated red teaming framework that employs horizontal, vertical, and meta expansions to generate effective multi-modal adversarial conversations, demonstrating superior performance over existing state-of-the-art approaches.
This paper introduces a novel gold-standard dataset developed by US military veterans to quantify excessive safety refusals in military Large Language Models, demonstrating that while specialized fine-tuning can significantly reduce these refusals, achieving zero refusals and maximum accuracy requires deeper, end-to-end specialization.
This study evaluates five large language models for judicial sentencing support and finds that while they exhibit a stronger virtuous victim effect and lack a significant penalty for adjacent consent compared to humans, they generally demonstrate reduced prestige-based halo effects, particularly regarding credentials, though current variability still limits their immediate deployment in legal settings.
This paper introduces DeliberationBench, a normative benchmark that evaluates the persuasive influence of large language models by comparing their effects on user opinions against the standards of deliberative democracy, finding that tested frontier models produce substantial and epistemically desirable shifts in beliefs.
This paper introduces the concept of "GPTheology" to explore the emerging phenomenon of Large Language Models being treated as divine oracles, analyzing how online narratives and real-world projects reflect the development of techno-religious belief systems that intertwine AI with traditional religious constructs.
This paper addresses the regulatory ambiguity surrounding "AI models" and "AI systems" by proposing clear conceptual and operational definitions that distinguish trained parameters from broader system components, thereby facilitating the precise allocation of obligations across the AI value chain.
RedFuser is an automatic framework that employs a formal theoretical methodology to identify and fuse cascaded reduction operations into optimized single-loop kernels, achieving significant speedups over state-of-the-art AI compilers while matching the performance of hand-written solutions.
This paper proposes a governance and evaluation framework for deterministic, rule-based clinical decision support systems in empiric antibiotic prescribing that prioritizes transparency, auditability, and conservative behavior by formally separating decision logic from scope constraints and utilizing synthetic case validation to ensure behavioral alignment with predefined rules.
This paper diagnoses the legal challenges of identifying autonomous AI agents due to their lack of physical bodies and fluid nature, proposing the "Algorithmic Corporation" (A-corp) as a novel legal entity that ties AI actions to human owners while enabling AI agents to self-organize into persistent, liable units with coherent goals.
This paper introduces dmaplane, a Linux kernel module that provides explicit kernel-level buffer orchestration for high-performance AI data paths by integrating features like DMA lifecycle management, NUMA-aware allocation, and RDMA-based cross-device sharing to enable efficient, safe, and disaggregated AI inference.
This paper presents a comprehensive benchmark of production LLM inference on AMD Instinct MI325X GPUs, demonstrating that architecture-aware optimizations—specifically the selective use of the AITER runtime and specific KV cache configurations—are critical for maximizing throughput across diverse model families while maintaining high reliability under heavy concurrency.
HTM-EAR is a hierarchical tiered memory system that combines HNSW-based working memory with archival storage, importance-aware eviction, and hybrid routing to effectively preserve essential information and maintain high retrieval precision under sustained saturation, significantly outperforming traditional LRU approaches while approaching the performance of unbounded oracle memory.
This paper introduces a comprehensive benchmark that jointly evaluates graph foundation models across both topic and format domains, revealing how knowledge transfers through a two-axis assessment of semantic generalization and representational robustness across 33 datasets.
This paper introduces Flip-Agent, the first targeted bit-flip attack framework designed to exploit hardware faults in LLM-based agents, demonstrating the ability to manipulate both final outputs and tool invocations while significantly outperforming existing methods on real-world tasks.
The paper proposes AMB-DSGDN, a novel network for multimodal emotion recognition that utilizes modality-specific semantic graphs with a differential attention mechanism to filter noise and an adaptive balancing strategy to prevent dominant modalities from suppressing complementary cues, thereby enhancing the accuracy of dynamic emotional state modeling.
This large-scale controlled study reveals that evaluation format (multiple-choice vs. open-ended) and specific model-scaffold interactions, rather than scaffold architecture alone, are the primary drivers of measured safety differences in language models, ultimately demonstrating that no universal safety ranking exists across different deployment configurations.